Professional Documents
Culture Documents
cellular networks?
prepared for
GSMA
05 May 2014
Disclaimer
1. Introduction................................................................................................. 2
6. Conclusion................................................................................................ 11
Appendices ...................................................................................................... 12
Appendix A: Incremental nature of Wi-Fi traffic ............................................ 12
Appendix B: Spectrum, traffic distribution and site numbers ........................ 14
Appendix C: Wi-Fi and cellular capacity solutions ........................................ 15
C.1. Hierarchy of capacity solutions ............................................................. 15
C.2. Cell range ............................................................................................. 15
C.3. Mobility and interworking ...................................................................... 16
C.4. Functionalities for Wi-Fi/Cellular integration.......................................... 16
Exhibits
Exhibit 1: Initiator of Wi-Fi Access .................................................................. 3
Exhibit 2: Traffic distribution ........................................................................... 4
Exhibit 3: Factors influencing technology choice for small cell capacity
solutions ......................................................................................... 7
Exhibit 4: Traffic forecasts according to Cisco and Ericsson ........................ 12
Exhibit 5: Distribution of traffic per site across network ................................ 14
Exhibit 6: Recently developed Wi-Fi and Cellular network integration
functions ....................................................................................... 17
Contact
0. Executive Summary
Wi-Fi provides an invaluable complement to cellular in the delivery of high quality
broadband services to smartphone users, particularly indoors. Cellular offers high
performance, wide area blanket coverage but does not always cover indoor locations
well. Wi-Fi fills these gaps at venues where local owners and users need improved
coverage and access speed. However, we see little evidence that today’s Wi-Fi
networks significantly reduce traffic levels on the cellular network. On the contrary,
there are signs that when high quality Wi-Fi and cellular are both available to users, the
traffic on both increases. Qualitative and quantitative analysis of some of the most
advanced markets reveals that the majority of today’s Wi-Fi traffic is incremental or
complementary to cellular traffic. Venue based Wi-Fi will play an increasing role in
providing indoor broadband connectivity. It will be used by cellular operators to deliver
an “always best connected” value proposition but will not reduce demand for capacity
on cellular networks.
While there are some outdoor Wi-Fi networks which provide connectivity in the
absence of sufficient 3G or LTE capacity, they are not an equivalent substitute for
building additional 3G and 4G capacity to cater for the surge in mobile broadband
traffic. We have identified a range of drawbacks to the Wi-Fi approach based on
technical, integration, availability, timing and operational factors. In order to deliver a
good mobile broadband experience, cellular operators require more LTE capacity.
Where necessary, mobile network operators will tend to use small LTE cells to relieve
congestion in traffic hotspots rather than relying on Wi-Fi.
The prospects for Wi-Fi delivering significant capacity relief in areas of the cellular
network facing congestion are limited. On the contrary, Wi-Fi and cellular traffic are
expected to grow in parallel and rapidly, offering complementary capabilities. Both
technologies will require additional spectrum to deliver ubiquitous broadband
connectivity.
1. Introduction
Wi-Fi today offers an invaluable Wi-Fi has proved to be immensely popular with smartphone users as a low cost
complement to cellular through low solution for improved localised coverage and mobile broadband experience, at venue
cost, venue-specific, broadband specific locations especially indoors. Industry sources state that between 60% or 80%
coverage. of traffic on a smartphone today is carried over Wi-Fi1. What is less evident is the
extent to which Wi-Fi can provide capacity relief to the cellular networks, either from
This paper addresses the potential of
existing Wi-Fi networks or potentially from new implementations such as small cells.
Wi-Fi also to reduce cellular capacity
With such high traffic volumes on Wi-Fi, it would be easy to infer that a high proportion
and spectrum needs.
of traffic is being offloaded from the cellular networks2, reducing the need for additional
cellular spectrum. But is this view valid?
This paper examines this question from a number of perspectives.
1. Is traffic carried over Wi-Fi incremental or replacement?
2. Do today’s coverage led Wi-Fi networks reduce cellular traffic load where it
matters?
3. How suitable is Wi-Fi as the base technology for small cell solution for cellular
capacity expansion?
4. What is the business motivation model for Wi-Fi networks to be built for
cellular capacity relief?
The term Wi-Fi Offload is used in a variety of ways by the industry. In this paper, we
use two different terms to distinguish the categories
Incremental/complementary. Wi-Fi traffic that, if it were not for the availability of Wi-
Fi access, would most probably not be generated at all. This could be for reasons
of cost, quality of service, applications restricted to Wi-Fi access. This type of traffic
is complementary to cellular and has a limited impact on cellular traffic load.
Replacement. Traffic that could equally well be carried over Wi-Fi or cellular but is
in fact carried over Wi-Fi for any reason. This reduces the traffic load on the cellular
network.
In the rest of the document, we use these two terms to qualify the different types of Wi-
Fi traffic.
The paper does not attempt to estimate spectrum needs for either cellular or Wi-Fi (see
for example ITU document3). Rather, the aim is to examine the long-term dynamic
between Wi-Fi and cellular usage, the suitability of Wi-Fi as a capacity solution, the
user and business motivation of Wi-Fi implementations and the resultant traffic loading
on the cellular network. It is traffic loading per site which ultimately drives cellular
spectrum needs.
building
Hyper-
dense
Office
Home
Retail
MAN
Initiator
In-
User
Device
Operator
Source: Coleago
Exhibit 1 shows in blue the most likely initiator of the Wi-Fi access. It is interesting to
note that the device is central to any Wi-Fi experience and that most of the venues
targeted by Wi-Fi networks are highly social venues.
Exhibit 2 also shows that whilst a large percentage of traffic is being carried on Wi-Fi
networks, the role of managed public Wi-Fi generally remains marginal in both
emerging and developed countries. Managed public Wi-Fi is a network that essentially
provides the same seamless experience as a cellular network and does not require any
manual authentication. It uses technologies such as EAP-SIM and Hotspot 2.0 which
are only now starting to be rolled out to any extent. Examples include London
Underground Wi-Fi and MNO provided Wi-Fi.
5% 2%
28%
32%
66%
67%
Cellular Private Wi-Fi Public Wi-Fi Cellular Private Wi-Fi Public Wi-Fi
Source: Mobidia - Understanding the Role of Managed Public Wi-Fi in Today’s Smartphone
User Experience (2012)
These venues are often inadequately serviced by cellular networks as the costs of
providing ubiquitous, deep indoor coverage for high quality broadband are prohibitive.
TCO analysis for an incremental GB of data on cellular vs Wi-Fi shows ratios
sometimes above 200% depending on the venue and the availability of backhaul. The
key difference is that the Wi-Fi coverage is highly localised and targeted, whereas
cellular coverage is wide area. Given the very high costs of wide-area coverage,
MNO’s have to prioritise investments in terms of locations covered and the depth of
coverage. This inevitably leaves gaps and MNO’s are interested in deploying Wi-Fi or
partnering with Wi-Fi network providers where the economics are justified.
The perception by the end user and their motivation for deliberately using a Wi-Fi
network is based on simple decision criteria:
Wi-Fi networks are free
Wi-Fi networks are fast, mostly because the perception is based on private Wi-Fi,
Wi-Fi networks are venue sensitive with bespoke marketing content.
Venue sensitivity may become important in the future with, for instance, McDonald’s
currently launching an app allowing its customers to order in the shop without placing
their order at the till.
Whether the smartphone is 2G, 3G or Traffic has been growing in both developed and developing markets. Forecasts from
4G, the share of Wi-Fi traffic above Ericsson4, Cisco5 and Informa6 show that the usage per smart device will grow
50% between 6 and 10 fold in the next 5 years. According to the analysis from Informa,
whilst some countries have moved from 3G to 4G, the share of Wi-Fi traffic as a
percentage of the total traffic does not deviate much from around 50% for Android
smartphone devices. This is despite the fact that LTE offers excellent broadband
experience at least in areas with good coverage. See Appendix A for further.
Home Wi-Fi is a good example of venues where users download GBytes of data every
month. Coverage from the cellular network in the home can be of insufficient quality for
a good broadband experience resulting in low usage. On the other hand, home Wi-Fi is
in effect free when the resident has chosen to sign up for a fixed line package. It should
be noted though that young adults often prefer not to have a fixed line subscription at
all. For household with a fixed line connection, there are no barriers to content and the
broadband experience is good; Wi-Fi access then becomes very attractive, driving very
high usage. This resultant traffic volume due to development of Wi-Fi networks is
largely incremental and does not cannibalise the cellular traffic but on the contrary
increases the overall data consumption.
The conclusion relating to usage is that Wi-Fi is socially rooted in the user’s
expectations of getting the best out of the local network from an availability, price,
application and traffic perspective. Wi-Fi networks are only available where the venue
owner has deployed Wi-Fi at its own cost as the MNO’s are still working up
monetisation strategies to push their Wi-Fi plans further. There is little evidence of an
arbitrage strategy between a cellular network and a Wi-Fi network. However, the
considerations above lead us to conclude that Wi-Fi traffic is mainly complementary to
cellular, not replacement. Whatever the precise categorisation and underlying drivers
of the different traffic types, what is certain is that traffic on Wi-Fi and cellular are both
growing rapidly and need to be served.
In city centres, high Wi-Fi availability (often from Public Wi-Fi networks) and high
cellular traffic levels often overlap. This creates opportunity for offload. However, this
overlap is not planned or consistent. Moreover, as seen in the previous section, Public
Wi-Fi accounts for a small fraction of total mobile traffic and the traffic is for the most
part incremental.
In residential areas, the most In more residential areas, Wi-Fi availability is present mainly in the form of home Wi-Fi.
significant form of Wi-Fi access is However, as argued in Section 2, the majority of Wi-Fi traffic generated in the home is
Home Wi-Fi. Most of this traffic is incremental to cellular constraining the offload opportunity. As for the potential use by
incremental but some will be genuine passing outdoor users, Home networks are generally private, blocking access to
replacement. passers-by. However commercial developments such as FON and Free Mobile
(France) enable authentication to Home Wi-Fi networks, albeit only to registered
subscribers. There are other obstacles for passing users of Home Wi-Fi such as the
limited coverage leakage from the indoor AP, and the handover from one AP to
another, or from an AP to the cellular network. These factors tend to limit the extent of
Wi-Fi offload in residential areas.
Some cities are deploying Metropolitan Area Networks (MANs), outdoor Wi-Fi networks
which on the face of it could provide valuable offload. They tend to be deployed in city
centre and business districts, mainly as an access alternative to cellular rather than as
a capacity solution. However, if the density of Wi-Fi sites were high enough, they could
achieve significant offload, relieving cellular traffic load. While there have been a
number of launches of such networks over the years, they remain untypical and have
had limited success to date.
In practice therefore, while the opportunity for Wi-Fi offload can arise in some public
spaces, there are many areas of high cellular traffic load where the Wi-Fi opportunity is
limited. This is even more pronounced in developing countries where the availability of
Wi-Fi and fixed broadband is often poor, especially beyond the city centre.
Exhibit 3: Factors influencing technology choice for small cell capacity solutions
Cell size / Power Traffic management OSS Interworking of Wi-Fi Widespread device
levels / cellular compatibility
Interference Mobility Policy management OSS across multiple
cancellation RATs
Security SON
Duration
Source: Coleago
Wi-Fi Spectrum
From an MNO perspective, a major concern of Wi-Fi spectrum is the fact that it is
shared with many other users and operators, usually unknown, and the traffic is
unmanaged or coordinated. In busy areas, this can lead to congestion in public areas,
manifesting itself as slower data rates or no connectivity. The opening up of the 5 GHz
spectrum bands with devices supporting them will help considerably to alleviate this
situation but rapid growth of Wi-Fi may still make this an issue in the longer term.
Traffic management
For MNOs, the ability to monitor traffic levels in the band, steer traffic between bands
and cells and implement policy rules are key requirements. Wi-Fi does not provide this
level of visibility and control, except for Wi-Fi deployed by the MNO itself.
Until recently, this was a significant gap with Wi-Fi but new developments such as
ANDSF (see Appendix C.3) are becoming available allowing better traffic monitoring
and policy control of traffic. This is primarily a client-based solution and a risk for the
MNO is that it is difficult to ensure that policy is being implemented consistently. The
feature would have to be widely supported on devices, demanding concerted effort and
widespread support from MNOs and vendors, backed by a certification regime for this
to happen. This could be challenging as MNO and vendor strategies on Wi-Fi vary.
Over the 5 to 10 year time-frame relevant to new spectrum, these technical solutions
should mature and stabilise. Given the huge diversity of Wi-Fi deployments, it may still
take a long time for these more advanced techniques to be deployed universally. Some
capabilities such as mobility, authentication, security mechanisms7 have been available
for years, some over 10 years, but have been slow to be implemented. For custom-
built, Wi-Fi small cell capacity solutions, the prospects for realising the required
functionality are much better, but still subject to the necessary device clients being
widely adopted.
Limited cell range of Wi-Fi
As cellular technologies, HSPA and LTE are designed to support large cells of many
km radius, whereas Wi-Fi is designed for low cost, short-range either indoors or
outdoors. Wi-Fi traffic is also moving towards the 5 GHz band which further reduces
the propagation range, although this may be offset by the use of MIMO technologies.
For outdoor small cell capacity solutions, the optimum size of the small cells will
depend on the capacity gain required, the level of in-building coverage, whether the
spectrum is dedicated or re-used, and the coverage area being targeted. In urban or
sub-urban areas where the macro layer cells are larger and a traffic more evenly
distributed, the area to be covered by small cells can be extensive. The flexibility
afforded by cellular technology regarding power levels and cell size allows both a more
effective solution (better coverage and capacity) and lower costs with fewer small
cells8.
Summary
As a potential technology for outdoor The main issues that we believe MNOs will have to contend with in selecting a base
capacity small cells, Wi-Fi has rf technology (Wi-Fi or HSPA/LTE) for small cells are:
performance, operational, timing and
functional draw-backs when Sharing of the Wi-Fi spectrum with multiple, uncoordinated users and the lack of
compared to HSPA or LTE ubiquitous traffic management.
Time taken for the necessary functionalities to mature and be widely available on
Wi-Fi networks and devices, recognising that small cell solutions are likely to be
needed in the near future. Better opportunities exist with purpose-built Wi-Fi small
cell layer off, but device compatibility remains a concern.
Operation, Configuration, Optimisation, SON and maintenance of Wi-Fi small cell
networks along side the main cellular network.
In the long run, all the telecommunication players are looking at providing ubiquitous
services and controlling the end-to-end user experience. MNO’s or MVNOs are likely to
set up wholesale agreements with Wi-Fi Network Providers combining their cellular
network services with Wi-Fi services to extend the reach and coverage, but not the
capacity of the cellular network.
The structure of the wholesale agreement between MNO’s and MSO’s may deter
MNOs from using Wi-Fi as a coverage or capacity solution. Wholesale agreements
being discussed today include wholesale Price per GByte, monthly Subscriber Active
Fee, monthly Fee per SSID activated per Access Point and any combinations of those.
Successful wholesale deals in Europe are generally based on a fixed fee per Wi-Fi AP
or per SSID; MNO’s are reluctant to sign deals based on a wholesale rate per GB.
6. Conclusion
The appeal of Wi-Fi today is rooted in venue specific coverage, high quality broadband,
and low cost. This model is proving to be highly successful with up to 70% of traffic on
smart-phones being carried over Wi-Fi. Notwithstanding this high level of Wi-Fi traffic,
cellular traffic is also growing rapidly and we do not see compelling evidence that Wi-Fi
traffic is replacing cellular traffic in a significant way. Instead, demand is growing for
both forms of connectivity. Furthermore, there is evidence that where good access to
both 4G and Wi-Fi are available, traffic over both carriers increases i.e. usage on one
stimulates usage on the other creating synergy between the two and a better overall
experience for the end user. We believe that going forward, this model will continue to
be successful, with Wi-Fi and cellular each delivering complementary parts of user’s
coverage needs. This will be further enhanced by innovative business models and
technical advances of the Wi-Fi technology.
In this context, the main question addressed in this paper is whether Wi-Fi can or will
alleviate capacity congestion on cellular networks. We considered first the offload
potential from venue specific, coverage-led implementations of Wi-Fi, the predominant
form of Wi-Fi today and probably the future; and secondly, the potential suitability of
Wi-Fi as the base technology for purpose-built capacity solutions in areas of cellular
congestion.
In locations where the cellular Existing Wi-Fi coverage-led networks. Cellular congestion occurs in many urban and
network is congested, Wi-Fi traffic suburban areas of the network, not just city centres. While opportunities for Wi-Fi
replacing cellular traffic is limited and offload exist in these critical areas, the amount of offload is small. The main reason is
patchy that Wi-Fi traffic is complementary but there are other constraints as well. Outside city
centres, Wi-Fi availability and the opportunity for cellular off-load are significantly
reduced, even in developed markets. The effect is even more pronounced in emerging
markets where Wi-Fi availability is often prevented due to an absence of fixed
broadband for backhaul.
Wi-Fi is a second choice to Wi-Fi as a capacity solution. Advances are being made with extending the functionality
HSPA/LTE as the preferred of Wi-Fi to match that of cellular which should in principle enable Wi-Fi based small cell
technology for deploying and solutions to be realised. However, there remain significant drawbacks to using Wi-Fi
operating small cells for capacity when compared with native cellular technology, which make small cells based on
HSPA or LTE a more effective and attractive proposition for many MNOs.
Notwithstanding these drawbacks, there will be circumstances where MNOs are unable
to deploy their own small cell solutions when carrier-grade Wi-Fi small networks could
provide a sensible solution.
We envisage that Wi-Fi will continue to grow rapidly delivering appealing, venue
specific, coverage-led solutions. Cellular and Wi-Fi forms of access are complementary
and the symbiotic existence drives greater demand of each. Both require new
spectrum allocations for the future.
Appendices
14,000
12,000
Monthly Usage in MB
10,000
8,000
6,000
4,000
2,000
0
2012 2013 2019
3G Smartphone (Ericsson) 3G Smartphone (Cisco) Laptop + Dongle (Ericsson)
Laptop + Dongle (Cisco) Tablet (Ericsson) Cisco - Tablet
Source: VNI Mobile Forecast Highlights, 2012 - 2017 - Cisco Systems - Ericsson Mobility
Report – Ericsson
If the Wi-Fi traffic were only driven by capacity offload, the Wi-Fi traffic coming from a
3G handset and a 4G handset would be expected to be roughly the same in absolute
terms. However, the analysis shows that almost twice as much traffic is carried over
Wi-Fi when using a 4G handset. The best way to look at this data is to assume that
70% of the time is spent by 3G or 4G users in areas where Wi-Fi networks are
available, and cellular networks are either limited in coverage or congested.
According to Localytics12, there is also a noticeable asymmetry in traffic being
offloaded by Apple 4G smartphones compared to iPad 4G tablets. Whilst the offloaded
traffic on an iphone 5 is close to 50%, the traffic offloaded on an iPad 4 is closer to
70%. As tablets usage would primarily be used in indoor venues, and due to the form
factor are based on a static experience, the difference in traffic patterns reinforces the
argument that Wi-Fi usage is driven by availability of the network rather than a
deliberate decision to choose one Wi-Fi bearer over a cellular one.
Confirming the coverage orientation of Wi-Fi traffic, a recent paper from Deutsche
Telecom13 showed that by comparing total traffic before and after introduction of Wi-Fi
in user trials, cellular traffic was unaffected by the introduction of free Wi-Fi access
The venues where Wi-Fi networks are available are unlikely to be intensively covered
by the mobile operators in the near future for the reasons explained earlier. Whilst the
composite traffic from a smart device will keep on growing, the proportion of Wi-Fi
offload should remain at around 60%-70%.
On the other hand, the challenge for companies rolling out Wi-Fi networks interested in
covering outdoor areas to a level on a par with that of cellular networks remains the
monetisation of such networks outside “real estate” venues.
In Coleago’s opinion, the status quo is likely to remain at the level seen today i.e. an
equilibrium determined by:
13 Wi-Fi Offloading.- Fairy tale or Swiss Army Knife? DT. Wi-Fi Offload Summit. January 2014.
8.0
Traffic per site relative to average
7.0
6.0
5.0
4.0
3.0
2.0
1.0
0.0
0%
Busiest - - - -20%
- - - - - - - - - 40% 60%- - - - - - - -80%
- - Network sites 100%
- - - Least busy
Source: Coleago
Most Wi-Fi traffic is today carried over the 2.4 GHz band which is rapidly becoming
congested in public and private places. Additional traffic will need to migrate to the 5
GHz band. Propagation in this band however is poorer than the lower bands making it
more suited to very short-range applications (this can be partly offset by MIMO). This
makes the band less suitable for outdoor Wi-Fi.
The link budget advantage of LTE and UMTS supports the use of bigger cells for the
small cell layer. This has to be balanced against the requirement that the cells must be
small enough to be replicated multiple times in order to re-use the available spectrum
and increase capacity. The advantage of the higher link budget of LTE (and UMTS) is
that it allows great flexibility of design to meet optimum trade-offs between
Cell size and coverage area
Wi-Fi is more constrained in this respect. This would be significant in busy areas of
cellular network where cell sizes are larger, such as in some urban and sub-urban
districts. In residential districts, the traffic is more evenly distributed and the small cells
will need to cover a relatively large area. With Wi-Fi, this will demand many more cells.
Standards body /
Category Standard / program Capability
organisation
Authentication and network discovery
WFA Hotspot 2.0 / Passpoint Facilitating and automating secure and trusted
Wi-Fi connectivity
Wi-Fi network discovery
IEEE 802.11u Building block of HotSpot 2.0
WBA NGH (Next generation
hotspot)
IETF EAP-AKA; EAP-SIM Secure authentication protocols
Network selection and traffic steering
3GPP ANDSF Client-based, policy driven control of network
selection and traffic steering
Being aligned with HotSpot2.0 functions
Mobility 3GPP SaMOG Mobility between 3GPP and Wi-Fi networks
Network integration 3GPP Trusted WLAN access Architecture giving Trusted WLAN access to
3GPP core (EPC). Based on SaMOG
Source: Various
These are important steps towards the full interworking needed to integrate Wi-Fi
capacity networks and cellular networks. Many of these are vital for Wi-Fi as a capacity
solution for cellular since the Wi-Fi access needs to be integrated and managed as a
seamless extension of the cellular network. The process of enhancing the Wi-Fi
capabilities continues.
What is now significant is that both cellular and Wi-Fi players are now pushing for
these new capabilities to be developed. The drivers come from the industry. If this
momentum is maintained over the 5 to 10 year time-frame relevant to new spectrum,
the prospects are good that the standards needed for effective Wi-Fi / cellular
interworking will be available. How ubiquitously they will be supported on devices
remains to be seen.